机载激光雷达点云的城市建筑物直线特征提取方法

Straight line feature extraction method of city airborne LiDAR point clouds

提出一种城市机载激光雷达点云的直线特征提取方法。首先基于曲率的区域增长方法分割三维点云平面,对每个点云平面进行二维投影,并进行网格化和二值化。其次通过稳健的二维线段检测方法获得平面的线特征,将二维线段重投影至三维空间得到相应的三维线段,构造线特征集。最后对线特征集构建基于固定网格的直线空间索引方法,实现基于线特征的点云索引。实验结果表明,该方法直线提取误差较低,相对平面相交法效率更高,能够有效提取城市建筑物的直线特征,通过索引提高了特征查询效率,可满足高精度区域分割和三维建模的需求。

This paper presents an algorithm for linear feature extraction of city airborne LiDAR point cloud.Firstly,the point cloud is segmented into 3Dplanes via region growing based on curvature,for each 3D planes.All the points belonging to it are projected onto the plane to form a 2Dimage,which is followed by gridded and binarized process.Secondly,the line segments of image obtained by a robust 2Dline segment detection method are then re-projected onto the 3Dplane to get the corresponding 3Dline segments and form the line feature set.Finally,a linear space index method based on fixed grid of the line feature set is proposed to realise the point cloud index based on linear feature.The experimental result shows that our algorithm has smaller error in linear extraction and higher efficiency than the plane intersection method.The algorithm can effectively extract the linear features of urban buildings,improve the efficiency of feature query through indexing,and meet the needs of high-precision region segmentation and 3Dmodeling.